Extending the concept of the (socio)linguistic variable to slang* Jannis K. Androutsopoulos (Heidelberg) [email protected] Published in Hungarian language in: Kis, Tamás (ed.) (2000). Mi a szleng? ['What is slang?'], pp. 109-140. Debrecen: Kossuth Lajos University Press. More information on this volume available at http://mnytud.arts.klte.hu/szleng/szl_kut/03miaszl/ Introduction It is the aim of this paper to propose an analytic scheme for the description of slang based on the concept of the (socio)linguistic variable. Based on data from a recent empirical study, I will argue that many slang items can be described in terms of linguistic variables, more precisely: as variants of specific types of variables. I will specify a number of slang variables on various levels of description, thereby discussing linguistic constraints which hold for each type of variable. A further focus of this paper is slang renewal, i.e. the creation of new slang variants. Overall, I will argue that an analysis of the kind proposed here makes it easier describe mechanisms of slang variation and change. On a theoretic level, this paper is an attempt at bridging the gap between slang studies and variationist sociolinguistics methods. It is no secret that slang has had no serious place in the agenda of sociolinguistic research, at best treated as a superficial vocabulary. Despite numerous dictionaries, slang still remains «an undersudied part of language» from a scholarly point of view (Eble 1998: 42). This fact is evident in the almost complete absence of slang treatments in major introductions to sociolinguistics. Hudson's statement, that slang «merits serious research by sociolinguists» (Hudson 1980: 53), is to a large extent still a desideratum. In fact, I would suggest that this neglect is also due to the way students of slang have * This is a considerably revised version of a paper entitled «Extending the concept of 'sociolinguistic variable' to German youth slang» which was first presented at the 1st Language Variation Workshop, held in April 1997 at the University of Reading, England. I am indebted to Janet Spreckels for improving my English. Jannis K. Androutsopoulos | Sociolinguistic variable and slang | -2- themselves approached their object of study. To be sure, recent literature includes both correlational approaches to slang (e.g. T. Labov 1982) and sociological ones (e.g. Donahue 1995), as well as papers which emphasize the need for an ethnographic approach to slang, in order to describe slang items in their discourse context (Szabó 1998). But as whole, traditional slang studies are mainly restricted to the lexicographic documentation, semantic classification, and etymological description of slang items, thereby almost completely ignoring the main methods and analytical tools of sociolinguistics. One neglected connection between slang studies and sociolinguistics seems to be the description of slang from a variationist point of view. Wolfram (1991: 22) points out that «all studies of linguistic variation operate on some notion of the linguistic variable». Indeed, many findings of slang studies are hard to conceive of without an implicit variable concept. It is my intention to do some explicit work with this concept, suggesting some ways it can be used for the description of slang. This paper's empirical base is data from contemporary German youth slang (Androutsopoulos 1998a). Young people's language, whatever its label may be, certainly is one of the most fruitful fields for slang studies today. Sornig (1990) considers young people in urban environments as one of the most important social carriers of slang today. Young people's speech is the data source for Slang and sociability (Eble 1996), one of the most important recent research monographs on slang. Furthermore, the only two mentions of «slang» in Sociolinguistic theory (Chambers 1995) deal with young people's speech. However, slang analysis is but one focus of contemporary research on youth language, other major foci being correlative approaches on the other hand, conversation-analytic and interactional sociolinguistic approaches on the other.1 The coordination among these research foci is still far from perfect. Correlative apopoaches on age-related linguistic variation have largely ignored youth slang (cf Chambers 1995:170-2), while qualitative approaches have equally ignored the discourse functions of slang items. This paper is not directly concernd with the question of «what slang is». My view is that «slang» is what linguists take it to be, and this does not always coincide with what lay speakers call «slang». Three aspects included in many definitions of slang2 are the restriction of slang on words and expressions, the restricted sociolinguistic distribution of slang, and its social function as a sign of in-groupness. While I will not deal with the latter two points, it is one aim of this paper to show the necessity to extend the descriptional range of slang analysis. That is, I will focus on the linguistic side of the issue, which explains my putting the 'socio' part of the (socio)linguistic variable in brackets. 1 See Androutsopoulos 1998b for a detailed review of international youth language research. 2 E.g. «informal, nonstandard vocabulary, usually intelligible only to people from a particular region or social group». (Crystal 1993, s.v. Slang); see also Andersson & Trudgill (1990: 69ff.); Eble (1996: 11); Eble (1998); Stenström (this volume). Jannis K. Androutsopoulos | Sociolinguistic variable and slang | -3- The (Socio)linguistic Variable Revisited It is not an overstatement to call the sociolinguistic variable one of the milestones of sociolinguistics. Using the linguistic variable as a structural unit and correlating it with social parameters is the core of the Labovian approach, which defined the discipline's first steps (Chambers 1995: 12-25). In general terms, a linguistic variable is a set of alternative ways of saying the same thing, to repeat Labov´s famous statement (Labov 1972, 94.). A sociolinguistic variable simply is a socially significant linguistic variable, in other words: a set of alternative linguistic realizations (variants) with social significance (Dittmar 1997: 57). A set of two or more realizations of an abstract linguistic unit, such as a phoneme, a grammatical function or a semantic unit, make up a linguistic variable. If these alternative realizations also carry different social meanings, each of them evoking a certain social or situational prototype, we are dealing with a sociolinguistic variable. Since the (socio)linguistic variable is not a frequently evoked concept in slang studies, it seems necessary to be a bit more specific about the criteria which must hold for a set of linguistic items to be a valid linguistic variable. I will do this by grouping the major points made in the literature3 under six headings, thereby keeping in mind that the linguistic features which form the bulk of slang, i.e. lexical units, are no prototypical sociolinguistic variables. Now a short discussion of these six properties in turn. Range of Application: Wolfram (1991) points out that the construct of the linguistic variable may apply on virtually all levels of linguistic analysis. He also explicitely aknowledges lexical choice as a type of linguistic variable. The prototypical sociolinguistic variables discussed in the literature are phonetic-phonological and to a certain extent also morphological ones. The respective status of syntactic structures, lexical items, and discourse markers is not always clear. The only kind of linguistic item Dittmar (1997:60) explicitely exludes from the sociolinguistic variable playground is the realisation of speech acts. As far as the lexicon is concerned, slang items are treated as variants by a number of authors. For instance, Halliday (1979) explicitly invokes variation theory for the description of «anti-language» (underworld argot) lexical items, while Armstrong (1998) works with traditional lexical pairs with a standard and a nonstandard variant.4 3 Discussion in this section is based on the following works: Albrecht (1986), Chambers (1995), Cheshire (1987), Dittmar (1997: 57-61), Fasold (1991), Fasold (1994: 223-4), Labov (1972), Lavandera (1978), Sankoff (1988), Schiffrin (1994: 341), Wolfram (1991). 4 Note also that in the classic version of implicational scale analysis as developed in creole studies, lexical variables (e.g. 'child' with one standard and one creole variant) are used next to phponological and morphological variables. Jannis K. Androutsopoulos | Sociolinguistic variable and slang | -4- Equivalence of Variants: Linguistic variants are «different ways to say the same thing», i.e. the variants which make out a variable are formally different, but their alternance should not in any way alter the «meaning» of the utterance. However, this condition seems more and more endangered the higher we go up the levels of linguistic description. Therefore, it is crucial to be precise about the two kinds of equivalence condition that are needed when working with slang data: (i) referential equivalence, i.e. two lexical items are variants if they have the same referential meaning, but a different expressive and/or social meaning. To use an English example, the items officer and cop are referencially equivalent in most contexts; (ii) functional equivalence, a notion introduced in correlational sociolinguistics by Lavandera (1978). Two items are functionally equivalent, if they have the